Download Design Tools: Human Body Modeling

Design Tools: Human Body
Modelling
Slide 2 of 41
Background
• Significant improvements in last thirty years due to
– Rapid developments in computer hardware and software
– Emphasis placed on the
• Reliable models of human body in an impact
• Numerous validation studies conducted using these models
• Economical and versatile method for the analyses of the crash
responses of complex dynamic systems.
• Applications:
– Reconstruction of actual accidents
– Computer aided design (CAD) of the crash response of vehicles, safety
devicesand roadside facilities and
– Human impact biomechanics studies.
Slide 3 of 41
Types of Models
• Deterministic:
The outcome of the crash event is predicted based
upon measured or estimated parameter values,
representing characteristics of the human body, safety
devices, the vehicle and its surroundings, using wellestablished physical laws.
• Statistical:
Used in injury biomechanics research to assess the
correct relationship between loading conditions and
resultant injuries by means of regression type of
equations (the so-called injury risk function).
Slide 4 of 41
Types of Models (Contd.)
• Although the various deterministic models may differ in
many aspects, all are dynamic models.
• The models account for inertial effects by deriving
equations of motions for all movable parts, and solving
these equations using an iterative method.
• The mathematical formulations used for these models can
be subdivided into
– lumped mass models
– multi-body models and
– finite element models.
• Lumped mass models are usually one- or two-dimensional,
multi-body models two- or three-dimensional and finite
element models are usually three-dimensional.
Slide 5 of 41
Lumped Mass Models
• The model consists of rigid bodies
with masses m1, m2 and m3
connected by springs and dampers.
• Mass m1 represents the Impactor
mass and masses m2 and m3 the
sternal and vertebral effective mass.
• Spring k12 represents the skin and
flesh between Impactor and sternum.
• The internal spring and dampers
represent the connection between
sternum and thoracic
• The response of this model was
shown to correlate well with human
cadaver tests
Figure from Priya Prasad, 2005
This model simulates the thorax response
in case of loading by an Impactor.
One-dimensional model of the human
thorax developed by Lobdell in 1973
Slide 6 of 41
Multi-Body Models
• Difference between a lumped mass model and a multi-body model :
– Elements in a multi-body formulation connected by various joint types
through which the number of degrees of freedom between the
elements can be constrained.
– A lumped mass model can be considered a special case of the more
general multi-body model formulation.
• The motion of the joint-connected elements in a multi-body model
is caused by external forces generated by so called force-interaction
models.
– Examples of force-interaction models in a multibody model for crash
analyses are the models to account for an acceleration field, springdamper elements, restraint system models and contact models.
• Another characteristic of lumped mass models is that in a multibody formulation, instead of rigid bodies, flexible bodies can be
specified.
Slide 7 of 41
Multi-Body Models
• Two-dimensional
model
of
human in restraints
• The human body part of the
model is characterized by rigid
bodies representing thorax/head,
upper arms, upper legs and lower
legs.
• Simple pin joints connect the rigid
elements.
• Able to show quite good
agreement
for
quantities
including hip displacements,
chest acceleration and belt loads.
Figure from Priya Prasad, 2005
Example of multi-body model: 7
degrees of freedom model for frontal
collisions by McHenry
Slide 8 of 41
Multi-Body Models
• Three-dimensional
MADYMO
model of a Chrysler Neon suitable
for frontal collisions
• Has more than 200 elements and
includes a description of interior,
restraint system, suspension,
steering wheel, bumper, engine
and hood
• 32-segment model of the Hybrid
III dummy
• Quite realistic results were
obtained when the model results
were compared with a rigid wall
test and offset deformable barrier
tests
Example of multi-body model: 3dimensional model for frontal collisions
of Chrysler Neon with Hybrid III dummy
*European Community and the United States’ National Highway Traffic Safety
Administration (NHTSA) to study vehicle compatibility issues
Figure from Priya Prasad, 2005
Slide 9 of 41
Finite Element Models
•
The system is divided into a number of
finite volumes, surfaces or lines
•
The state of stress follows deformations
and
the
constitutive
material
properties.
•
The model was developed in the 70s by
Shugar.
– Represents the skull and brain.
– Linear elastic and linear viscoelastic material behavior was
assumed.
– Skull bone response and brain
response
compared
with
experimental results of head
impact tests with primates.
Figure from Priya Prasad, 2005
A finite element model simulating the
human body: a head model by Shugar
Slide 10 of 41
Multi body Vs. Finite Element
Multibody
• Very efficient way complex
kinematic connections as
present in the human body
and in parts of the vehicle
structure like the steering
assembly and the vehicle
suspension system
Finite element
• Capability of describing (local)
structural deformations and
stress distribution
• Study of injury mechanisms in
the human body parts.
• Usually longer computer times
• Finite element method less
attractive for optimization
studies involving many design
parameters
Slide 11 of 41
Hybrid approach
• Airbag (and airbag straps)
– modeled in PISCES 3D-ELK program
(now MSC-DYTRAN) using almost
2000 triangular membrane elements.
– Perfect gas law.
– Accounts leakage through airbag
material and exhaust orifices
– Pressure and temperature constant
– Inertia effects of the gas neglected
• FE airbag with a multi-body
Hybrid III dummy in MADYMO
Figure from Priya Prasad, 2005
An integrated multi-body finite element
model: occupant-airbag interaction by
Bruijs
Slide 12 of 41
Outline
• The theoretical basis
– The Multibody Method for Crash Analyses.
– Integrated Multi-body Finite Element Simulations
• Human body models for crash analyses
– Crash Dummy Modeling
– Real human body models
• Conclusion, future trends are discussed
Slide 13 of 41
Multi-Body Method for Crash Analyses
MADYMO setup:
• The multi-body module of
the program calculates the
contribution of the inertia
of bodies to the equations
of motion.
• Special models are available
for
vehicle
dynamic
applications including tire
models.
• A control module– Apply forces and Torques
– Sensors to receive signals
from bodies.
Figure from Priya Prasad, 2005
MADYMO modules
Slide 14 of 41
Topology of a System of Bodies
•
•
•
•
•
For topology specification the chains have
to be reduced to a tree structure.
Closed chains of bodies permitted in the
later MADYMO versions.
Reference body given number 1.
The other bodies are numbered from 2 to
N on the path from the reference body to
any other body are lower than the
number of that specific body.
In the MADYMO input file, the
configuration of a system is defined by
entering for each branch the numbers of
the bodies in decreasing order.
Figure from Priya Prasad, 2005
Branch 1: 3 2 1
Branch 2: 7 2 1
Branch 3: 6 5 1
Branch 4: 8 4 1
Slide 15 of 41
Specification of motion of a rigid Body
• A right hand base {e}i .
• Origin chosen co-incident with
the center of mass
• Newton-Euler equations
• Position of the origin and the
orientation of the body-fixed
base relative to an inertial
base {E}.
• Position of the origin of the
body-fixed base relative to the
origin of the inertial base is
given by the vector ri.
• Orientation relative to the
inertial base is defined by the
rotation matrix Ai.
Figure from Priya Prasad, 2005
Slide 16 of 41
Kinematics of a Flexible Body
• Bodies that experience small deformations can be
modeled as flexible
• Motion of a point on a body
– Rigid body motion
– Superimposed motion due to the deformation
• Motions due to deformations approximated by a linear
combination of predefined displacement and rotation
fields (deformation modes).
• Only at certain pre-defined points in the body (the
nodes), the deformation nodes are defined
Kinematics of a Pair of Bodies
Connected by a Joint
• Kinematic joint as defined
here can connect only two
bodies
• In MADYMO, motion of a
body ‘j’ is described relative
to the corresponding lower
numbered body ‘i’.
– ωi and ωj are the angular
velocity of body ‘I’ and ‘j’,
respectively, and
– ωij the angular velocity of the
joint.
Figure from MADYMO manual
Slide 17 of 41
Slide 18 of 41
Some Kinematic joints
Translational-rotational
joint
Figure from MADYMO manual
Slide 19 of 41
Equations of Motion
The constraint forces and torques
can be eliminated using the
principle of virtual work
• Ýi is a 6 × 1 column matrix that
contains the components of the
linear and angular acceleration of
the base of the lower numbered
body i.
• Matrices Mij and Qij are
calculated successively, starting
with body N to body 1
• This algorithm yields the second
time derivatives of the joint
coordinates in explicit form
• Two explicit numerical integration
methods
– Fourth order Runge- Kutta method
that uses a constant time step;
– Fifth order Runge-Kutta- Merson
method which uses a variable time
step that is controlled by the local
truncation error.
Slide 20 of 41
Force Interaction Models
The motion of a system of joint-connected bodies is
caused by applied forces. The models in MADYMO are:
• Acceleration field model
• Spring-damper elements
• Muscle models
• Contact models
• Belt model
• Dynamic joint models
*Allows custom codes
Slide 21 of 41
Acceleration field model
• Calculates the forces at the
centers of gravity of bodies
in a homogeneous timedependent
acceleration
field “a” .
• If the vehicle does not
rotate during the crash, the
actual
recorded
accelerations at the vehicle
can be prescribed as an
acceleration field acting on
the occupant, while the
vehicle is connected to the
inertial space.
Figure from Priya Prasad, 2005
A homogeneous acceleration field
Slide 22 of 41
Spring-damper elements
• The Kelvin - a uniaxial element
spring parallel with a damper.
• The Maxwell - a uniaxial element
spring and damper in series.
• The point-restraint model combination of three Kelvin
elements with infinite spring
length, each parallel to one of the
axes of an orthogonal coordinate
system.
• All
spring-damper
models
attached to arbitrary points of any
two bodies or between a body and
the inertial space.
Figure from MADYMO manual
Slide 23 of 41
Muscle models
•
Most common muscle model in
biomechanical research is the Hill model :
–
Contractile Element (CE) : active force
generated by the muscle
–
Parallel Elastic element (PE) : elastic
properties of muscle fibers and
surrounding tissue
–
Elastic Elements (SE1 and SE2) : elastic
properties of tendons and aponeurosis
–
(M1 and M2) : Muscle mass.
•
Basic muscle model in MADYMO consists
of the CE and the PE.
•
Muscles with varying complexity can be
formulated using this basic model in
combination with the standard MADYMO
elements.
Figure from Priya Prasad, 2005
The Hill-type muscle model
Slide 24 of 41
Modeling Muscle Contraction - Hill Model
Force-Length Relation
 Muscle generate max. force at
Lopt
 Force decreases either side of
Lopt
 Values of Fmax and Lopt are
taken from Delp et al. (1990)
 Muscle force
increasing Vmax
Force-Velocity Relation
decreases
with
 Vmax: Function of fraction of fast
fibers in a muscle (the more the fast
fibers, the more the value of Vmax).
Fraction values for each muscle are
taken from Yamaguchi et al. (1990)
Activation Level
Hill muscle model
CE = Contractile component to account for the active muscle
forces
PE & DE = Viscoelastic components (Spring & Damper) to account
for passive muscle forces
Total Muscle Force = FCE + FPE + FDE , FCE = Na(t) * Fl(x) * Fv(v)
(Na max= 1.0)
Na int A
B
D’
C
(0.005)
Ttrig Tref = 20 ms
D
Slower
Muscle
Faster
Muscle
Time
Muscle State – time Relation
 Tref. =20 ms (Ackerman 2002)
 Activation time constant (Ta):
function of fraction of slow fibers
in a muscle (the more the slow
fibers, the more the time needed
to reach max. activation level).
Figure from Priya Prasad, 2005
Slide 25 of 41
Contact model
• Arbitrary shaped surfaces (called
facet surfaces) are used to model
contact with other bodies or the
surroundings.
• Ellipse of order and still higher
order ‘n’ can be described for
bodies
•
•
•
•
Surfaces cannot deform ,instead
allowed to penetrate into each other.
A contact force is generated between
two colliding surfaces as a function of
the penetration of the two surfaces
as well as of the relative velocity in
the contact area
Elastic (including hysteresis and
dynamic amplification), damping and
friction forces can be specified in the
contact.
If a facet surface is involved, the
contact force, instead of being “forcepenetration”–based, may be based
on a “stress-penetration” function.
Contact loads in an ellipsoid-ellipsoid
contact (only forces acting on the upper
ellipsoid are shown)
Slide 26 of 41
The Belt Model
• Belt model :a chain of
connected, mass-less, springtype segments
• End points connected to rigid
bodies or the inertial space
“attachment points”
• Attachment
points
cannot
change during a simulation
• A webbing-sensitive reel locks if
the belt feed rate exceeds a
specified limit.
• A pre-tensioner model is also
available.
Figure from Priya Prasad, 2005
A 3-point belt with retractor
Slide 27 of 41
Dynamic Joint Models
•
Two types of loads :
–
–
•
•
The internal forces and torques caused by
the kinematic joint constraints
The applied loads : passive loads due to
friction or elastic resistance or active loads
caused by muscle activity.
Simplest type generates a force(or torque)
as function of single joint coordinate.
Complicated “flexion-torsion restraint”
–
–
–
Applied to spherical joints in flexible
structures
Relative joint position of the joint is
considered to be the result of two
successive rotations
Bending stiffness in forward direction can
differ from the stiffness in backward or
lateral bending.
Figure from Priya Prasad, 2005
A revolute joint with joint torques
ex. Neck and spine in crash dummies.
Slide 28 of 41
Integrated Multi-Body Finite Element
Simulations
•
•
•
•
Elements implemented : truss, beam,
shell, brick and membrane
material models : elastic, viscoelastic, elastoplastic, hysteresis and
Moonley-Rivlin
Models also are available : sandwich
material, solid foam and honeycomb
A MADYMO model can be
– multi-body systems, finite element
structures or combinations of the two.
•
MADYMO and finite element
programs like PAMCRASH, LSDYNA,
RADIOSS and MSC DYTRAN can be
externally coupled.
Interaction between multi-body and
finite element module
ex. The airbag model can be attached
to the multi-body steering column
using supports while the interaction
of the airbag with a multi-body
occupant can be handled through
contacts.
Slide 29 of 41
Crash Dummy Modelling
Need for mechanical models of human:
• Requirement from design departments in the
automotive industry, for well-validated design
tools, which can reduce the number of
regulatory tests with crash dummies
• Crash dummy models can be measured with
relative ease unlike biological models
Slide 30 of 41
Modelling Methodology
Divide into
segments
• As per functional areas
• Part has significant mass and flexible joints
• Flexible structures modeled by two universal joints
• Outer surface by ellipsoids
Inertial Properties
• Complete Inertia tensor
• The stiffness of the connections (joints)
• Dependency on more than one degree of freedom is
neglected
Surface
compliance
properties
• Static as well as dynamic measurements with several
penetrating surface
• Skin covering thickness and density
Slide 31 of 41
Examples of Crash Dummy Databases
(1/2)
Hybrid III 3yr, Hybrid III 6yr and
Crabi 12 month and TNO child
dummies P3/4, P3, P6 and P10)
Figure from Priya Prasad, 2005
Hybrid III dummy family (top) and 5th 50th
and 95%, USA child dummies (bottom)
Slide 32 of 41
Examples of Crash Dummy Databases
(2/2)
FACET
Model
Figure from Priya Prasad, 2005
Finite
Element
Model
Modelling the Real Human
Body
• Difficult to develop than a model of physical
crash dummy
• Mathematical modeling of the real human
body offers improved “Biofidelity” compared
to crash dummy models
– study of aspects like body size, body posture,
muscular activity and post fracture response
• Detailed human body models potentially allow
analysis of injury mechanisms on a material
level
• A large number of models describing specific
parts of the human body have been published
but only a few of these models describe the
response of the entire human body in impact
conditions
Slide 33 of 41
Slide 34 of 41
Anthropometry (1/2)
• GEBOD is often used to generate models
representing arbitrary human body sizes
• GEBOD generates a model consisting of 15
segments: head, neck, upper arms, lower arms,
thorax, abdomen, pelvis, upper legs, lower legs
and feet
• The parameters generated by GEBOD using
regression equations on the basis of body height
and weight for adult males and females and along
with age combination for children
Slide 35 of 41
Anthropometry (2/2)
•
•
•
•
•
RAMSIS developed for ergonomic
analyses
The RAMSIS model :the human body
as a set of rigid bodies connected by
kinematic joints and the skin is
described as a triangulated surface.
Segment mass and center of gravity
are derived in RAMSIS using this
realistic geometric description.
Provides a mathematical prediction
for the increase of the average body
height of the entire population
during a given time period (secular
growth).
Describes the entire population in a
realistic way
Figure from Priya Prasad, 2005
MADYMO human models of various body
sizes generated from the RAMSIS model,
from left to right: 3-year-old child,
extremely small female, 50th percentile
male, extremely large male
Slide 36 of 41
Examples of a Human Body Model
(1/3)
• Sled test at conducted by the Naval
Biodynamics Laboratory in New Orleans
• Neck - a global 7-segment model is available
with lumped properties
Figure from Priya Prasad, 2005
Slide 37 of 41
Examples of a Human Body Model
(2/3)
MADYMO finite element brain model
developed at the Eindhoven Technical
University, The Netherlands
Figure from Priya Prasad, 2005
Comparison of human neck model response
and volunteer response
Slide 38 of 41
Examples of a Human Body Model
(3/3)
• Represents a 50th-percentile male
• Developed by Lizee et al in the
RADIOSS program package
• Detailed representations of the
neck, shoulder, thorax and pelvis
• Model has been validated in more
than 30 test configurations.
• The model has more than 10,000
elements.
• Head, arms and legs (not shown)
represented as rigid bodies.
Figure from Priya Prasad, 2005
Finite element human body model
developed by Lizee et. al.[
Slide 39 of 41
Conclusion (1/4)
• Major advantage of the multibody approach:
– Simulate in an efficient way
– Spatial motions of mechanical systems
– Complex kinematic connections in the human body and in parts
of the vehicle structure
• Advantage of the finite element method : Describing local
deformations and stresses in a realistic way.
• Creation of a finite element model is time-consuming, and the
availability of realistic material data is limited, particularly in the case
of biological tissue response
• Large amounts of computer time hence method less attractive for
complex optimization studies involving many design parameters
Slide 40 of 41
Conclusion (2/4)
• Availability of well-validated databases of the human body is an
important condition for the use of Computer crash models
• Promising results achieved in real human body models in the nineties
• A unique advantage of a design strategy based on real human body
crash models over a design strategy based on crash tests with
dummies is the possibility to rapidly benefit from new scientific
knowledge of injury mechanisms and injury criteria obtained through
biomechanical research
• New scientific findings have seldom resulted in improvements in the
dummy design particularly since safety regulations, which specify the
Hybrid III dummy as a regulatory test device, tend to freeze the
specifications in the regulation for a long period
Slide 41 of 41
Conclusion (3/4)
• An increased usage of computer models also can be
observed in the area of accident reconstruction and
litigation
• Areas of future development in the field of real human
body models include further improvements in
– the description of the non-linear dynamic behavior of
muscles
– the modeling of complex human joints and
– study of constitutive equations and parameters for
biological materials such as the brain and skin
Slide 42 of 41
Conclusion (4/4)
• The usage of finite element techniques
coupled with multi-body techniques will allow
the user to benefit from the capabilities of
both approaches and will offer the flexibility
of merging more global multi-body models
with,
whenever
needed,
detailed
representations for certain parts in his model.